End of training

README.md

@@ -3,6 +3,8 @@ library_name: transformers
 license: apache-2.0
 base_model: timm/resnet18.a1_in1k
 tags:
+- image-classification
+- vision
 - generated_from_trainer
 metrics:
 - accuracy
@@ -16,7 +18,7 @@ should probably proofread and complete it, then remove this comment. -->
 
 # vit-base-beans
 
-This model is a fine-tuned version of [timm/resnet18.a1_in1k](https://huggingface.co/timm/resnet18.a1_in1k) on an unknown dataset.
+This model is a fine-tuned version of [timm/resnet18.a1_in1k](https://huggingface.co/timm/resnet18.a1_in1k) on the beans dataset.
 It achieves the following results on the evaluation set:
 - Loss: 0.7389
 - Accuracy: 0.8045

all_results.json

@@ -1,13 +1,13 @@
 {
     "epoch": 15.0,
-    "eval_accuracy": 0.7894736842105263,
-    "eval_loss": 0.8550169467926025,
-    "eval_runtime": 0.7658,
-    "eval_samples_per_second": 173.669,
-    "eval_steps_per_second": 22.198,
+    "eval_accuracy": 0.8045112781954887,
+    "eval_loss": 0.7388833165168762,
+    "eval_runtime": 0.856,
+    "eval_samples_per_second": 155.372,
+    "eval_steps_per_second": 19.86,
     "total_flos": 1.5658365504595968e+17,
-    "train_loss": 0.6299933981284117,
-    "train_runtime": 92.772,
-    "train_samples_per_second": 167.184,
-    "train_steps_per_second": 21.019
+    "train_loss": 0.923483537771763,
+    "train_runtime": 151.8082,
+    "train_samples_per_second": 102.168,
+    "train_steps_per_second": 12.845
 }

eval_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 15.0,
-    "eval_accuracy": 0.7894736842105263,
-    "eval_loss": 0.8550169467926025,
-    "eval_runtime": 0.7658,
-    "eval_samples_per_second": 173.669,
-    "eval_steps_per_second": 22.198
+    "eval_accuracy": 0.8045112781954887,
+    "eval_loss": 0.7388833165168762,
+    "eval_runtime": 0.856,
+    "eval_samples_per_second": 155.372,
+    "eval_steps_per_second": 19.86
 }

runs/Nov05_18-37-10_ip-10-90-1-182/events.out.tfevents.1730831994.ip-10-90-1-182.2476579.1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9a5ac7adbfdb9f91ce2c30406e678fcd7d233d78c1800655312a0c129f6660d1
+size 411

train_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 15.0,
     "total_flos": 1.5658365504595968e+17,
-    "train_loss": 0.6299933981284117,
-    "train_runtime": 92.772,
-    "train_samples_per_second": 167.184,
-    "train_steps_per_second": 21.019
+    "train_loss": 0.923483537771763,
+    "train_runtime": 151.8082,
+    "train_samples_per_second": 102.168,
+    "train_steps_per_second": 12.845
 }

trainer_state.json

@@ -1,5 +1,5 @@
 {
-  "best_metric": 0.8550169467926025,
+  "best_metric": 0.7388833165168762,
   "best_model_checkpoint": "./beans_outputs/checkpoint-1950",
   "epoch": 15.0,
   "eval_steps": 500,
@@ -10,1512 +10,1512 @@
   "log_history": [
     {
       "epoch": 0.07692307692307693,
-      "grad_norm": 2.0955963134765625,
-      "learning_rate": 1.9692307692307696e-05,
+      "grad_norm": 2.1216182708740234,
+      "learning_rate": 1.98974358974359e-05,
       "loss": 1.1239,
       "step": 10
     },
     {
       "epoch": 0.15384615384615385,
-      "grad_norm": 1.8236017227172852,
-      "learning_rate": 1.9384615384615386e-05,
-      "loss": 1.1222,
+      "grad_norm": 1.8308407068252563,
+      "learning_rate": 1.9794871794871798e-05,
+      "loss": 1.1221,
       "step": 20
     },
     {
       "epoch": 0.23076923076923078,
-      "grad_norm": 1.9866633415222168,
-      "learning_rate": 1.907692307692308e-05,
-      "loss": 1.1165,
+      "grad_norm": 1.9811038970947266,
+      "learning_rate": 1.9692307692307696e-05,
+      "loss": 1.1163,
       "step": 30
     },
     {
       "epoch": 0.3076923076923077,
-      "grad_norm": 2.4556338787078857,
-      "learning_rate": 1.876923076923077e-05,
-      "loss": 1.1047,
+      "grad_norm": 2.4690616130828857,
+      "learning_rate": 1.958974358974359e-05,
+      "loss": 1.1046,
       "step": 40
     },
     {
       "epoch": 0.38461538461538464,
-      "grad_norm": 1.6562598943710327,
-      "learning_rate": 1.8461538461538465e-05,
-      "loss": 1.1085,
+      "grad_norm": 1.6778465509414673,
+      "learning_rate": 1.9487179487179488e-05,
+      "loss": 1.1082,
       "step": 50
     },
     {
       "epoch": 0.46153846153846156,
-      "grad_norm": 1.501336932182312,
-      "learning_rate": 1.8153846153846155e-05,
-      "loss": 1.1048,
+      "grad_norm": 1.4839682579040527,
+      "learning_rate": 1.9384615384615386e-05,
+      "loss": 1.1043,
       "step": 60
     },
     {
       "epoch": 0.5384615384615384,
-      "grad_norm": 2.606107234954834,
-      "learning_rate": 1.784615384615385e-05,
-      "loss": 1.1031,
+      "grad_norm": 2.6265501976013184,
+      "learning_rate": 1.9282051282051284e-05,
+      "loss": 1.1028,
       "step": 70
     },
     {
       "epoch": 0.6153846153846154,
-      "grad_norm": 2.634584426879883,
-      "learning_rate": 1.753846153846154e-05,
-      "loss": 1.0916,
+      "grad_norm": 2.565593719482422,
+      "learning_rate": 1.9179487179487182e-05,
+      "loss": 1.0909,
       "step": 80
     },
     {
       "epoch": 0.6923076923076923,
-      "grad_norm": 2.039886236190796,
-      "learning_rate": 1.7230769230769234e-05,
-      "loss": 1.095,
+      "grad_norm": 2.020240545272827,
+      "learning_rate": 1.907692307692308e-05,
+      "loss": 1.0939,
       "step": 90
     },
     {
       "epoch": 0.7692307692307693,
-      "grad_norm": 2.4800949096679688,
-      "learning_rate": 1.6923076923076924e-05,
-      "loss": 1.0916,
+      "grad_norm": 2.49810791015625,
+      "learning_rate": 1.8974358974358975e-05,
+      "loss": 1.0917,
       "step": 100
     },
     {
       "epoch": 0.8461538461538461,
-      "grad_norm": 1.6766040325164795,
-      "learning_rate": 1.6615384615384618e-05,
-      "loss": 1.0959,
+      "grad_norm": 1.8923050165176392,
+      "learning_rate": 1.8871794871794873e-05,
+      "loss": 1.0937,
       "step": 110
     },
     {
       "epoch": 0.9230769230769231,
-      "grad_norm": 2.2848353385925293,
-      "learning_rate": 1.630769230769231e-05,
-      "loss": 1.1005,
+      "grad_norm": 2.2869341373443604,
+      "learning_rate": 1.876923076923077e-05,
+      "loss": 1.1001,
       "step": 120
     },
     {
       "epoch": 1.0,
-      "grad_norm": 4.07199239730835,
-      "learning_rate": 1.6000000000000003e-05,
-      "loss": 1.0881,
+      "grad_norm": 4.069129467010498,
+      "learning_rate": 1.866666666666667e-05,
+      "loss": 1.0863,
       "step": 130
     },
     {
       "epoch": 1.0,
-      "eval_accuracy": 0.41353383458646614,
-      "eval_loss": 1.0901767015457153,
-      "eval_runtime": 0.7894,
-      "eval_samples_per_second": 168.48,
-      "eval_steps_per_second": 21.535,
+      "eval_accuracy": 0.42857142857142855,
+      "eval_loss": 1.0881553888320923,
+      "eval_runtime": 0.9817,
+      "eval_samples_per_second": 135.479,
+      "eval_steps_per_second": 17.317,
       "step": 130
     },
     {
       "epoch": 1.0769230769230769,
-      "grad_norm": 2.0711560249328613,
-      "learning_rate": 1.5692307692307693e-05,
-      "loss": 1.0794,
+      "grad_norm": 1.7320090532302856,
+      "learning_rate": 1.8564102564102567e-05,
+      "loss": 1.0792,
       "step": 140
     },
     {
       "epoch": 1.1538461538461537,
-      "grad_norm": 2.499732494354248,
-      "learning_rate": 1.5384615384615387e-05,
-      "loss": 1.0755,
+      "grad_norm": 2.5494987964630127,
+      "learning_rate": 1.8461538461538465e-05,
+      "loss": 1.072,
       "step": 150
     },
     {
       "epoch": 1.2307692307692308,
-      "grad_norm": 1.4569323062896729,
-      "learning_rate": 1.5076923076923078e-05,
-      "loss": 1.0807,
+      "grad_norm": 1.4740803241729736,
+      "learning_rate": 1.835897435897436e-05,
+      "loss": 1.0773,
       "step": 160
     },
     {
       "epoch": 1.3076923076923077,
-      "grad_norm": 2.351478338241577,
-      "learning_rate": 1.4769230769230772e-05,
-      "loss": 1.0965,
+      "grad_norm": 2.3478026390075684,
+      "learning_rate": 1.8256410256410257e-05,
+      "loss": 1.0907,
       "step": 170
     },
     {
       "epoch": 1.3846153846153846,
-      "grad_norm": 2.1514322757720947,
-      "learning_rate": 1.4461538461538462e-05,
-      "loss": 1.0839,
+      "grad_norm": 2.1786012649536133,
+      "learning_rate": 1.8153846153846155e-05,
+      "loss": 1.08,
       "step": 180
     },
     {
       "epoch": 1.4615384615384617,
-      "grad_norm": 2.151601791381836,
-      "learning_rate": 1.4153846153846156e-05,
-      "loss": 1.0837,
+      "grad_norm": 2.1593339443206787,
+      "learning_rate": 1.8051282051282053e-05,
+      "loss": 1.0789,
       "step": 190
     },
     {
       "epoch": 1.5384615384615383,
-      "grad_norm": 1.797500729560852,
-      "learning_rate": 1.3846153846153847e-05,
-      "loss": 1.0807,
+      "grad_norm": 2.1174795627593994,
+      "learning_rate": 1.794871794871795e-05,
+      "loss": 1.0779,
       "step": 200
     },
     {
       "epoch": 1.6153846153846154,
-      "grad_norm": 1.7583892345428467,
-      "learning_rate": 1.353846153846154e-05,
-      "loss": 1.073,
+      "grad_norm": 1.8251662254333496,
+      "learning_rate": 1.784615384615385e-05,
+      "loss": 1.067,
       "step": 210
     },
     {
       "epoch": 1.6923076923076923,
-      "grad_norm": 3.5006496906280518,
-      "learning_rate": 1.3230769230769231e-05,
-      "loss": 1.0741,
+      "grad_norm": 3.535820245742798,
+      "learning_rate": 1.7743589743589744e-05,
+      "loss": 1.0701,
       "step": 220
     },
     {
       "epoch": 1.7692307692307692,
-      "grad_norm": 2.0179672241210938,
-      "learning_rate": 1.2923076923076925e-05,
-      "loss": 1.077,
+      "grad_norm": 2.0242528915405273,
+      "learning_rate": 1.7641025641025642e-05,
+      "loss": 1.0714,
       "step": 230
     },
     {
       "epoch": 1.8461538461538463,
-      "grad_norm": 2.057086944580078,
-      "learning_rate": 1.2615384615384616e-05,
-      "loss": 1.0925,
+      "grad_norm": 2.1979262828826904,
+      "learning_rate": 1.753846153846154e-05,
+      "loss": 1.0868,
       "step": 240
     },
     {
       "epoch": 1.9230769230769231,
-      "grad_norm": 1.9710103273391724,
-      "learning_rate": 1.230769230769231e-05,
-      "loss": 1.0757,
+      "grad_norm": 2.096327543258667,
+      "learning_rate": 1.7435897435897438e-05,
+      "loss": 1.0682,
       "step": 250
     },
     {
       "epoch": 2.0,
-      "grad_norm": 2.9315028190612793,
-      "learning_rate": 1.2e-05,
-      "loss": 1.0716,
+      "grad_norm": 3.008599042892456,
+      "learning_rate": 1.7333333333333336e-05,
+      "loss": 1.063,
       "step": 260
     },
     {
       "epoch": 2.0,
-      "eval_accuracy": 0.5037593984962406,
-      "eval_loss": 1.0685255527496338,
-      "eval_runtime": 0.7711,
-      "eval_samples_per_second": 172.471,
-      "eval_steps_per_second": 22.045,
+      "eval_accuracy": 0.5413533834586466,
+      "eval_loss": 1.0589852333068848,
+      "eval_runtime": 0.7866,
+      "eval_samples_per_second": 169.088,
+      "eval_steps_per_second": 21.613,
       "step": 260
     },
     {
       "epoch": 2.076923076923077,
-      "grad_norm": 2.183527946472168,
-      "learning_rate": 1.1692307692307694e-05,
-      "loss": 1.0786,
+      "grad_norm": 2.03446888923645,
+      "learning_rate": 1.7230769230769234e-05,
+      "loss": 1.0712,
       "step": 270
     },
     {
       "epoch": 2.1538461538461537,
-      "grad_norm": 2.379652500152588,
-      "learning_rate": 1.1384615384615385e-05,
-      "loss": 1.0676,
+      "grad_norm": 2.4047350883483887,
+      "learning_rate": 1.7128205128205128e-05,
+      "loss": 1.0593,
       "step": 280
     },
     {
       "epoch": 2.230769230769231,
-      "grad_norm": 2.387296438217163,
-      "learning_rate": 1.1076923076923079e-05,
-      "loss": 1.064,
+      "grad_norm": 2.37992262840271,
+      "learning_rate": 1.7025641025641026e-05,
+      "loss": 1.0577,
       "step": 290
     },
     {
       "epoch": 2.3076923076923075,
-      "grad_norm": 2.8164947032928467,
-      "learning_rate": 1.076923076923077e-05,
-      "loss": 1.0548,
+      "grad_norm": 2.940575361251831,
+      "learning_rate": 1.6923076923076924e-05,
+      "loss": 1.044,
       "step": 300
     },
     {
       "epoch": 2.3846153846153846,
-      "grad_norm": 1.849363088607788,
-      "learning_rate": 1.0461538461538463e-05,
-      "loss": 1.0693,
+      "grad_norm": 1.983780026435852,
+      "learning_rate": 1.6820512820512822e-05,
+      "loss": 1.0573,
       "step": 310
     },
     {
       "epoch": 2.4615384615384617,
-      "grad_norm": 2.0274300575256348,
-      "learning_rate": 1.0153846153846154e-05,
-      "loss": 1.0604,
+      "grad_norm": 1.9737600088119507,
+      "learning_rate": 1.671794871794872e-05,
+      "loss": 1.048,
       "step": 320
     },
     {
       "epoch": 2.5384615384615383,
-      "grad_norm": 1.4544223546981812,
-      "learning_rate": 9.846153846153848e-06,
-      "loss": 1.0458,
+      "grad_norm": 1.4561364650726318,
+      "learning_rate": 1.6615384615384618e-05,
+      "loss": 1.0317,
       "step": 330
     },
     {
       "epoch": 2.6153846153846154,
-      "grad_norm": 2.1712183952331543,
-      "learning_rate": 9.53846153846154e-06,
-      "loss": 1.0609,
+      "grad_norm": 2.187842845916748,
+      "learning_rate": 1.6512820512820513e-05,
+      "loss": 1.0482,
       "step": 340
     },
     {
       "epoch": 2.6923076923076925,
-      "grad_norm": 1.922677993774414,
-      "learning_rate": 9.230769230769232e-06,
-      "loss": 1.0646,
+      "grad_norm": 1.9264109134674072,
+      "learning_rate": 1.641025641025641e-05,
+      "loss": 1.0491,
       "step": 350
     },
     {
       "epoch": 2.769230769230769,
-      "grad_norm": 1.8288860321044922,
-      "learning_rate": 8.923076923076925e-06,
-      "loss": 1.0597,
+      "grad_norm": 1.8492430448532104,
+      "learning_rate": 1.630769230769231e-05,
+      "loss": 1.0409,
       "step": 360
     },
     {
       "epoch": 2.8461538461538463,
-      "grad_norm": 2.123480796813965,
-      "learning_rate": 8.615384615384617e-06,
-      "loss": 1.052,
+      "grad_norm": 2.1464972496032715,
+      "learning_rate": 1.6205128205128207e-05,
+      "loss": 1.0359,
       "step": 370
     },
     {
       "epoch": 2.9230769230769234,
-      "grad_norm": 1.820168137550354,
-      "learning_rate": 8.307692307692309e-06,
-      "loss": 1.0434,
+      "grad_norm": 1.8184483051300049,
+      "learning_rate": 1.6102564102564105e-05,
+      "loss": 1.0237,
       "step": 380
     },
     {
       "epoch": 3.0,
-      "grad_norm": 4.973505020141602,
-      "learning_rate": 8.000000000000001e-06,
-      "loss": 1.061,
+      "grad_norm": 4.10977029800415,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 1.0447,
       "step": 390
     },
     {
       "epoch": 3.0,
-      "eval_accuracy": 0.6240601503759399,
-      "eval_loss": 1.0459144115447998,
-      "eval_runtime": 0.7618,
-      "eval_samples_per_second": 174.588,
-      "eval_steps_per_second": 22.316,
+      "eval_accuracy": 0.6992481203007519,
+      "eval_loss": 1.022922396659851,
+      "eval_runtime": 0.831,
+      "eval_samples_per_second": 160.057,
+      "eval_steps_per_second": 20.458,
       "step": 390
     },
     {
       "epoch": 3.076923076923077,
-      "grad_norm": 2.1673223972320557,
-      "learning_rate": 7.692307692307694e-06,
-      "loss": 1.0616,
+      "grad_norm": 2.4164669513702393,
+      "learning_rate": 1.5897435897435897e-05,
+      "loss": 1.0417,
       "step": 400
     },
     {
       "epoch": 3.1538461538461537,
-      "grad_norm": 1.8567888736724854,
-      "learning_rate": 7.384615384615386e-06,
-      "loss": 1.0501,
+      "grad_norm": 1.9611194133758545,
+      "learning_rate": 1.5794871794871795e-05,
+      "loss": 1.03,
       "step": 410
     },
     {
       "epoch": 3.230769230769231,
-      "grad_norm": 2.0640571117401123,
-      "learning_rate": 7.076923076923078e-06,
-      "loss": 1.0611,
+      "grad_norm": 2.1152236461639404,
+      "learning_rate": 1.5692307692307693e-05,
+      "loss": 1.038,
       "step": 420
     },
     {
       "epoch": 3.3076923076923075,
-      "grad_norm": 2.215384006500244,
-      "learning_rate": 6.76923076923077e-06,
-      "loss": 1.0469,
+      "grad_norm": 2.2656948566436768,
+      "learning_rate": 1.558974358974359e-05,
+      "loss": 1.0163,
       "step": 430
     },
     {
       "epoch": 3.3846153846153846,
-      "grad_norm": 2.1000049114227295,
-      "learning_rate": 6.461538461538463e-06,
-      "loss": 1.0499,
+      "grad_norm": 2.1608102321624756,
+      "learning_rate": 1.548717948717949e-05,
+      "loss": 1.0277,
       "step": 440
     },
     {
       "epoch": 3.4615384615384617,
-      "grad_norm": 1.7218382358551025,
-      "learning_rate": 6.153846153846155e-06,
-      "loss": 1.0564,
+      "grad_norm": 1.7433905601501465,
+      "learning_rate": 1.5384615384615387e-05,
+      "loss": 1.0379,
       "step": 450
     },
     {
       "epoch": 3.5384615384615383,
-      "grad_norm": 2.3569300174713135,
-      "learning_rate": 5.846153846153847e-06,
-      "loss": 1.0599,
+      "grad_norm": 2.625889778137207,
+      "learning_rate": 1.5282051282051282e-05,
+      "loss": 1.0314,
       "step": 460
     },
     {
       "epoch": 3.6153846153846154,
-      "grad_norm": 1.5210909843444824,
-      "learning_rate": 5.538461538461539e-06,
-      "loss": 1.0367,
+      "grad_norm": 1.6532872915267944,
+      "learning_rate": 1.517948717948718e-05,
+      "loss": 1.0052,
       "step": 470
     },
     {
       "epoch": 3.6923076923076925,
-      "grad_norm": 2.7621657848358154,
-      "learning_rate": 5.230769230769232e-06,
-      "loss": 1.0421,
+      "grad_norm": 2.715625524520874,
+      "learning_rate": 1.5076923076923078e-05,
+      "loss": 1.0101,
       "step": 480
     },
     {
       "epoch": 3.769230769230769,
-      "grad_norm": 1.5097808837890625,
-      "learning_rate": 4.923076923076924e-06,
-      "loss": 1.0362,
+      "grad_norm": 1.6237256526947021,
+      "learning_rate": 1.4974358974358976e-05,
+      "loss": 0.9981,
       "step": 490
     },
     {
       "epoch": 3.8461538461538463,
-      "grad_norm": 1.5118447542190552,
-      "learning_rate": 4.615384615384616e-06,
-      "loss": 1.0572,
+      "grad_norm": 1.582321286201477,
+      "learning_rate": 1.4871794871794874e-05,
+      "loss": 1.0226,
       "step": 500
     },
     {
       "epoch": 3.9230769230769234,
-      "grad_norm": 1.7513490915298462,
-      "learning_rate": 4.307692307692308e-06,
-      "loss": 1.0361,
+      "grad_norm": 1.7656166553497314,
+      "learning_rate": 1.4769230769230772e-05,
+      "loss": 0.9989,
       "step": 510
     },
     {
       "epoch": 4.0,
-      "grad_norm": 5.398025035858154,
-      "learning_rate": 4.000000000000001e-06,
-      "loss": 1.0514,
+      "grad_norm": 5.592857837677002,
+      "learning_rate": 1.4666666666666666e-05,
+      "loss": 1.0223,
       "step": 520
     },
     {
       "epoch": 4.0,
-      "eval_accuracy": 0.6015037593984962,
-      "eval_loss": 1.0407124757766724,
-      "eval_runtime": 0.7726,
-      "eval_samples_per_second": 172.155,
-      "eval_steps_per_second": 22.005,
+      "eval_accuracy": 0.6917293233082706,
+      "eval_loss": 0.9968266487121582,
+      "eval_runtime": 0.8013,
+      "eval_samples_per_second": 165.971,
+      "eval_steps_per_second": 21.214,
       "step": 520
     },
     {
       "epoch": 4.076923076923077,
-      "grad_norm": 2.5516345500946045,
-      "learning_rate": 3.692307692307693e-06,
-      "loss": 1.0529,
+      "grad_norm": 2.7708559036254883,
+      "learning_rate": 1.4564102564102564e-05,
+      "loss": 1.0136,
       "step": 530
     },
     {
       "epoch": 4.153846153846154,
-      "grad_norm": 1.6976008415222168,
-      "learning_rate": 3.384615384615385e-06,
-      "loss": 1.0472,
+      "grad_norm": 1.880313515663147,
+      "learning_rate": 1.4461538461538462e-05,
+      "loss": 1.0166,
       "step": 540
     },
     {
       "epoch": 4.230769230769231,
-      "grad_norm": 2.5672519207000732,
-      "learning_rate": 3.0769230769230774e-06,
-      "loss": 1.0565,
+      "grad_norm": 2.6722023487091064,
+      "learning_rate": 1.435897435897436e-05,
+      "loss": 1.0238,
       "step": 550
     },
     {
       "epoch": 4.3076923076923075,
-      "grad_norm": 2.166529655456543,
-      "learning_rate": 2.7692307692307697e-06,
-      "loss": 1.0619,
+      "grad_norm": 2.3504159450531006,
+      "learning_rate": 1.4256410256410258e-05,
+      "loss": 1.0338,
       "step": 560
     },
     {
       "epoch": 4.384615384615385,
-      "grad_norm": 1.961472511291504,
-      "learning_rate": 2.461538461538462e-06,
-      "loss": 1.0322,
+      "grad_norm": 2.2466416358947754,
+      "learning_rate": 1.4153846153846156e-05,
+      "loss": 0.9838,
       "step": 570
     },
     {
       "epoch": 4.461538461538462,
-      "grad_norm": 2.392319440841675,
-      "learning_rate": 2.153846153846154e-06,
-      "loss": 1.0388,
+      "grad_norm": 2.590592861175537,
+      "learning_rate": 1.405128205128205e-05,
+      "loss": 0.9915,
       "step": 580
     },
     {
       "epoch": 4.538461538461538,
-      "grad_norm": 2.3034205436706543,
-      "learning_rate": 1.8461538461538465e-06,
-      "loss": 1.0358,
+      "grad_norm": 2.3943469524383545,
+      "learning_rate": 1.3948717948717949e-05,
+      "loss": 0.9885,
       "step": 590
     },
     {
       "epoch": 4.615384615384615,
-      "grad_norm": 2.037050247192383,
-      "learning_rate": 1.5384615384615387e-06,
-      "loss": 1.0334,
+      "grad_norm": 1.8981530666351318,
+      "learning_rate": 1.3846153846153847e-05,
+      "loss": 0.9802,
       "step": 600
     },
     {
       "epoch": 4.6923076923076925,
-      "grad_norm": 3.0737335681915283,
-      "learning_rate": 1.230769230769231e-06,
-      "loss": 1.0506,
+      "grad_norm": 3.2020835876464844,
+      "learning_rate": 1.3743589743589745e-05,
+      "loss": 0.9993,
       "step": 610
     },
     {
       "epoch": 4.769230769230769,
-      "grad_norm": 2.1824796199798584,
-      "learning_rate": 9.230769230769232e-07,
-      "loss": 1.0516,
+      "grad_norm": 2.171642541885376,
+      "learning_rate": 1.3641025641025643e-05,
+      "loss": 1.0033,
       "step": 620
     },
     {
       "epoch": 4.846153846153846,
-      "grad_norm": 1.9239214658737183,
-      "learning_rate": 6.153846153846155e-07,
-      "loss": 1.0399,
+      "grad_norm": 1.8594719171524048,
+      "learning_rate": 1.353846153846154e-05,
+      "loss": 0.9689,
       "step": 630
     },
     {
       "epoch": 4.923076923076923,
-      "grad_norm": 2.267302989959717,
-      "learning_rate": 3.0769230769230774e-07,
-      "loss": 1.0374,
+      "grad_norm": 2.3601503372192383,
+      "learning_rate": 1.3435897435897435e-05,
+      "loss": 0.9703,
       "step": 640
     },
     {
       "epoch": 5.0,
-      "grad_norm": 4.4957404136657715,
-      "learning_rate": 0.0,
-      "loss": 1.05,
+      "grad_norm": 5.1482744216918945,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 1.0,
       "step": 650
     },
     {
       "epoch": 5.0,
-      "eval_accuracy": 0.6766917293233082,
-      "eval_loss": 1.0332472324371338,
-      "eval_runtime": 0.8208,
-      "eval_samples_per_second": 162.035,
-      "eval_steps_per_second": 20.711,
+      "eval_accuracy": 0.7518796992481203,
+      "eval_loss": 0.9575085639953613,
+      "eval_runtime": 0.8114,
+      "eval_samples_per_second": 163.92,
+      "eval_steps_per_second": 20.952,
       "step": 650
     },
     {
       "epoch": 5.076923076923077,
-      "grad_norm": 1.9310355186462402,
+      "grad_norm": 1.8753620386123657,
       "learning_rate": 1.3230769230769231e-05,
-      "loss": 1.0281,
+      "loss": 0.9664,
       "step": 660
     },
     {
       "epoch": 5.153846153846154,
-      "grad_norm": 2.36603045463562,
+      "grad_norm": 2.668313503265381,
       "learning_rate": 1.312820512820513e-05,
-      "loss": 1.0587,
+      "loss": 1.0055,
       "step": 670
     },
     {
       "epoch": 5.230769230769231,
-      "grad_norm": 2.0758917331695557,
+      "grad_norm": 2.544369697570801,
       "learning_rate": 1.3025641025641027e-05,
-      "loss": 1.0534,
+      "loss": 0.9936,
       "step": 680
     },
     {
       "epoch": 5.3076923076923075,
-      "grad_norm": 2.549725294113159,
+      "grad_norm": 2.7921175956726074,
       "learning_rate": 1.2923076923076925e-05,
-      "loss": 1.0421,
+      "loss": 0.9852,
       "step": 690
     },
     {
       "epoch": 5.384615384615385,
-      "grad_norm": 2.7183680534362793,
+      "grad_norm": 2.6936609745025635,
       "learning_rate": 1.2820512820512823e-05,
-      "loss": 1.035,
+      "loss": 0.9725,
       "step": 700
     },
     {
       "epoch": 5.461538461538462,
-      "grad_norm": 1.7176955938339233,
+      "grad_norm": 2.4392247200012207,
       "learning_rate": 1.2717948717948718e-05,
-      "loss": 1.0268,
+      "loss": 0.9653,
       "step": 710
     },
     {
       "epoch": 5.538461538461538,
-      "grad_norm": 1.8257861137390137,
+      "grad_norm": 1.9628013372421265,
       "learning_rate": 1.2615384615384616e-05,
-      "loss": 1.0302,
+      "loss": 0.9497,
       "step": 720
     },
     {
       "epoch": 5.615384615384615,
-      "grad_norm": 2.50368595123291,
+      "grad_norm": 2.3558847904205322,
       "learning_rate": 1.2512820512820514e-05,
-      "loss": 1.0331,
+      "loss": 0.9646,
       "step": 730
     },
     {
       "epoch": 5.6923076923076925,
-      "grad_norm": 2.4315121173858643,
+      "grad_norm": 2.4973325729370117,
       "learning_rate": 1.2410256410256412e-05,
-      "loss": 1.0208,
+      "loss": 0.9458,
       "step": 740
     },
     {
       "epoch": 5.769230769230769,
-      "grad_norm": 2.043854236602783,
+      "grad_norm": 2.091280221939087,
       "learning_rate": 1.230769230769231e-05,
-      "loss": 1.0241,
+      "loss": 0.9506,
       "step": 750
     },
     {
       "epoch": 5.846153846153846,
-      "grad_norm": 2.0800740718841553,
+      "grad_norm": 2.1584551334381104,
       "learning_rate": 1.2205128205128208e-05,
-      "loss": 1.0251,
+      "loss": 0.9517,
       "step": 760
     },
     {
       "epoch": 5.923076923076923,
-      "grad_norm": 3.4356396198272705,
+      "grad_norm": 3.8231847286224365,
       "learning_rate": 1.2102564102564102e-05,
-      "loss": 1.0104,
+      "loss": 0.9375,
       "step": 770
     },
     {
       "epoch": 6.0,
-      "grad_norm": 4.49416971206665,
+      "grad_norm": 4.746103286743164,
       "learning_rate": 1.2e-05,
-      "loss": 1.0357,
+      "loss": 0.9726,
       "step": 780
     },
     {
       "epoch": 6.0,
-      "eval_accuracy": 0.6541353383458647,
-      "eval_loss": 1.0109117031097412,
-      "eval_runtime": 0.7801,
-      "eval_samples_per_second": 170.483,
-      "eval_steps_per_second": 21.791,
+      "eval_accuracy": 0.7744360902255639,
+      "eval_loss": 0.92984938621521,
+      "eval_runtime": 0.8125,
+      "eval_samples_per_second": 163.686,
+      "eval_steps_per_second": 20.922,
       "step": 780
     },
     {
       "epoch": 6.076923076923077,
-      "grad_norm": 2.650513172149658,
+      "grad_norm": 2.8764736652374268,
       "learning_rate": 1.1897435897435898e-05,
-      "loss": 1.0148,
+      "loss": 0.9424,
       "step": 790
     },
     {
       "epoch": 6.153846153846154,
-      "grad_norm": 2.3682632446289062,
+      "grad_norm": 2.512113332748413,
       "learning_rate": 1.1794871794871796e-05,
-      "loss": 1.0094,
+      "loss": 0.9378,
       "step": 800
     },
     {
       "epoch": 6.230769230769231,
-      "grad_norm": 1.6716077327728271,
+      "grad_norm": 1.8056284189224243,
       "learning_rate": 1.1692307692307694e-05,
-      "loss": 1.0097,
+      "loss": 0.9309,
       "step": 810
     },
     {
       "epoch": 6.3076923076923075,
-      "grad_norm": 2.4839890003204346,
+      "grad_norm": 2.3125550746917725,
       "learning_rate": 1.1589743589743592e-05,
-      "loss": 1.01,
+      "loss": 0.9316,
       "step": 820
     },
     {
       "epoch": 6.384615384615385,
-      "grad_norm": 2.4004769325256348,
+      "grad_norm": 2.485017776489258,
       "learning_rate": 1.1487179487179487e-05,
-      "loss": 1.0104,
+      "loss": 0.9359,
       "step": 830
     },
     {
       "epoch": 6.461538461538462,
-      "grad_norm": 2.9597084522247314,
+      "grad_norm": 3.3460640907287598,
       "learning_rate": 1.1384615384615385e-05,
-      "loss": 1.0137,
+      "loss": 0.9316,
       "step": 840
     },
     {
       "epoch": 6.538461538461538,
-      "grad_norm": 2.680335760116577,
+      "grad_norm": 2.9212265014648438,
       "learning_rate": 1.1282051282051283e-05,
-      "loss": 1.0147,
+      "loss": 0.9417,
       "step": 850
     },
     {
       "epoch": 6.615384615384615,
-      "grad_norm": 1.7677160501480103,
+      "grad_norm": 1.9060611724853516,
       "learning_rate": 1.117948717948718e-05,
-      "loss": 0.9947,
+      "loss": 0.9127,
       "step": 860
     },
     {
       "epoch": 6.6923076923076925,
-      "grad_norm": 2.0484132766723633,
+      "grad_norm": 2.729116201400757,
       "learning_rate": 1.1076923076923079e-05,
-      "loss": 1.0169,
+      "loss": 0.9581,
       "step": 870
     },
     {
       "epoch": 6.769230769230769,
-      "grad_norm": 2.1910479068756104,
+      "grad_norm": 2.170494794845581,
       "learning_rate": 1.0974358974358977e-05,
-      "loss": 1.0024,
+      "loss": 0.9267,
       "step": 880
     },
     {
       "epoch": 6.846153846153846,
-      "grad_norm": 2.181236743927002,
+      "grad_norm": 2.36336350440979,
       "learning_rate": 1.0871794871794871e-05,
-      "loss": 0.9962,
+      "loss": 0.9209,
       "step": 890
     },
     {
       "epoch": 6.923076923076923,
-      "grad_norm": 2.898885488510132,
+      "grad_norm": 3.067629098892212,
       "learning_rate": 1.076923076923077e-05,
-      "loss": 1.012,
+      "loss": 0.9425,
       "step": 900
     },
     {
       "epoch": 7.0,
-      "grad_norm": 4.950052738189697,
+      "grad_norm": 4.193312168121338,
       "learning_rate": 1.0666666666666667e-05,
-      "loss": 1.0012,
+      "loss": 0.9258,
       "step": 910
     },
     {
       "epoch": 7.0,
-      "eval_accuracy": 0.7368421052631579,
-      "eval_loss": 0.981479287147522,
-      "eval_runtime": 0.7693,
-      "eval_samples_per_second": 172.886,
-      "eval_steps_per_second": 22.098,
+      "eval_accuracy": 0.8045112781954887,
+      "eval_loss": 0.8871035575866699,
+      "eval_runtime": 0.9311,
+      "eval_samples_per_second": 142.845,
+      "eval_steps_per_second": 18.258,
       "step": 910
     },
     {
       "epoch": 7.076923076923077,
-      "grad_norm": 2.692753553390503,
+      "grad_norm": 2.743016004562378,
       "learning_rate": 1.0564102564102565e-05,
-      "loss": 0.9889,
+      "loss": 0.9111,
       "step": 920
     },
     {
       "epoch": 7.153846153846154,
-      "grad_norm": 2.9175124168395996,
+      "grad_norm": 3.4307682514190674,
       "learning_rate": 1.0461538461538463e-05,
-      "loss": 0.9911,
+      "loss": 0.9108,
       "step": 930
     },
     {
       "epoch": 7.230769230769231,
-      "grad_norm": 3.221527099609375,
+      "grad_norm": 3.426872968673706,
       "learning_rate": 1.0358974358974361e-05,
-      "loss": 0.9827,
+      "loss": 0.9011,
       "step": 940
     },
     {
       "epoch": 7.3076923076923075,
-      "grad_norm": 2.507923126220703,
+      "grad_norm": 2.8081014156341553,
       "learning_rate": 1.0256410256410256e-05,
-      "loss": 0.9919,
+      "loss": 0.9076,
       "step": 950
     },
     {
       "epoch": 7.384615384615385,
-      "grad_norm": 2.4533870220184326,
+      "grad_norm": 2.9387893676757812,
       "learning_rate": 1.0153846153846154e-05,
-      "loss": 0.9962,
+      "loss": 0.9236,
       "step": 960
     },
     {
       "epoch": 7.461538461538462,
-      "grad_norm": 2.1032631397247314,
+      "grad_norm": 2.192082643508911,
       "learning_rate": 1.0051282051282052e-05,
-      "loss": 0.9751,
+      "loss": 0.8889,
       "step": 970
     },
     {
       "epoch": 7.538461538461538,
-      "grad_norm": 2.5848186016082764,
+      "grad_norm": 2.60426926612854,
       "learning_rate": 9.94871794871795e-06,
-      "loss": 0.9939,
+      "loss": 0.9243,
       "step": 980
     },
     {
       "epoch": 7.615384615384615,
-      "grad_norm": 2.17742919921875,
+      "grad_norm": 2.2395219802856445,
       "learning_rate": 9.846153846153848e-06,
-      "loss": 0.9745,
+      "loss": 0.8859,
       "step": 990
     },
     {
       "epoch": 7.6923076923076925,
-      "grad_norm": 1.9953967332839966,
+      "grad_norm": 2.0519468784332275,
       "learning_rate": 9.743589743589744e-06,
-      "loss": 0.9665,
+      "loss": 0.8821,
       "step": 1000
     },
     {
       "epoch": 7.769230769230769,
-      "grad_norm": 3.0263218879699707,
+      "grad_norm": 3.1931142807006836,
       "learning_rate": 9.641025641025642e-06,
-      "loss": 0.9646,
+      "loss": 0.8825,
       "step": 1010
     },
     {
       "epoch": 7.846153846153846,
-      "grad_norm": 2.3735406398773193,
+      "grad_norm": 2.4902703762054443,
       "learning_rate": 9.53846153846154e-06,
-      "loss": 0.9836,
+      "loss": 0.9106,
       "step": 1020
     },
     {
       "epoch": 7.923076923076923,
-      "grad_norm": 2.548480272293091,
+      "grad_norm": 2.7248635292053223,
       "learning_rate": 9.435897435897436e-06,
-      "loss": 0.9546,
+      "loss": 0.8597,
       "step": 1030
     },
     {
       "epoch": 8.0,
-      "grad_norm": 3.8450028896331787,
+      "grad_norm": 4.114623069763184,
       "learning_rate": 9.333333333333334e-06,
-      "loss": 0.9932,
+      "loss": 0.9203,
       "step": 1040
     },
     {
       "epoch": 8.0,
-      "eval_accuracy": 0.7669172932330827,
-      "eval_loss": 0.9549766778945923,
-      "eval_runtime": 0.7536,
-      "eval_samples_per_second": 176.486,
-      "eval_steps_per_second": 22.558,
+      "eval_accuracy": 0.8345864661654135,
+      "eval_loss": 0.8486866354942322,
+      "eval_runtime": 0.8033,
+      "eval_samples_per_second": 165.565,
+      "eval_steps_per_second": 21.162,
       "step": 1040
     },
     {
       "epoch": 8.076923076923077,
-      "grad_norm": 1.6870847940444946,
+      "grad_norm": 1.7978770732879639,
       "learning_rate": 9.230769230769232e-06,
-      "loss": 0.9752,
+      "loss": 0.8922,
       "step": 1050
     },
     {
       "epoch": 8.153846153846153,
-      "grad_norm": 2.1122217178344727,
+      "grad_norm": 2.1388049125671387,
       "learning_rate": 9.128205128205129e-06,
-      "loss": 0.9573,
+      "loss": 0.8696,
       "step": 1060
     },
     {
       "epoch": 8.23076923076923,
-      "grad_norm": 4.1552886962890625,
+      "grad_norm": 3.935715913772583,
       "learning_rate": 9.025641025641027e-06,
-      "loss": 0.9764,
+      "loss": 0.9006,
       "step": 1070
     },
     {
       "epoch": 8.307692307692308,
-      "grad_norm": 1.7864203453063965,
+      "grad_norm": 2.129542112350464,
       "learning_rate": 8.923076923076925e-06,
-      "loss": 0.9434,
+      "loss": 0.8603,
       "step": 1080
     },
     {
       "epoch": 8.384615384615385,
-      "grad_norm": 2.2091946601867676,
+      "grad_norm": 2.4084651470184326,
       "learning_rate": 8.820512820512821e-06,
-      "loss": 0.974,
+      "loss": 0.8989,
       "step": 1090
     },
     {
       "epoch": 8.461538461538462,
-      "grad_norm": 2.4063644409179688,
+      "grad_norm": 2.7610435485839844,
       "learning_rate": 8.717948717948719e-06,
-      "loss": 0.9576,
+      "loss": 0.8732,
       "step": 1100
     },
     {
       "epoch": 8.538461538461538,
-      "grad_norm": 1.6061931848526,
+      "grad_norm": 1.6848160028457642,
       "learning_rate": 8.615384615384617e-06,
-      "loss": 0.9588,
+      "loss": 0.8715,
       "step": 1110
     },
     {
       "epoch": 8.615384615384615,
-      "grad_norm": 2.8999595642089844,
+      "grad_norm": 3.277689218521118,
       "learning_rate": 8.512820512820513e-06,
-      "loss": 0.9791,
+      "loss": 0.8989,
       "step": 1120
     },
     {
       "epoch": 8.692307692307692,
-      "grad_norm": 3.6554131507873535,
+      "grad_norm": 3.7184839248657227,
       "learning_rate": 8.410256410256411e-06,
-      "loss": 0.9629,
+      "loss": 0.8822,
       "step": 1130
     },
     {
       "epoch": 8.76923076923077,
-      "grad_norm": 1.7246966361999512,
+      "grad_norm": 2.3165555000305176,
       "learning_rate": 8.307692307692309e-06,
-      "loss": 0.9707,
+      "loss": 0.889,
       "step": 1140
     },
     {
       "epoch": 8.846153846153847,
-      "grad_norm": 2.3160033226013184,
+      "grad_norm": 2.629028558731079,
       "learning_rate": 8.205128205128205e-06,
-      "loss": 0.9578,
+      "loss": 0.8753,
       "step": 1150
     },
     {
       "epoch": 8.923076923076923,
-      "grad_norm": 2.685718059539795,
+      "grad_norm": 2.8620731830596924,
       "learning_rate": 8.102564102564103e-06,
-      "loss": 0.972,
+      "loss": 0.8855,
       "step": 1160
     },
     {
       "epoch": 9.0,
-      "grad_norm": 3.6465442180633545,
+      "grad_norm": 3.924490451812744,
       "learning_rate": 8.000000000000001e-06,
-      "loss": 0.9748,
+      "loss": 0.9038,
       "step": 1170
     },
     {
       "epoch": 9.0,
-      "eval_accuracy": 0.7669172932330827,
-      "eval_loss": 0.9408761858940125,
-      "eval_runtime": 0.7615,
-      "eval_samples_per_second": 174.644,
-      "eval_steps_per_second": 22.323,
+      "eval_accuracy": 0.8120300751879699,
+      "eval_loss": 0.8329855799674988,
+      "eval_runtime": 0.781,
+      "eval_samples_per_second": 170.303,
+      "eval_steps_per_second": 21.768,
       "step": 1170
     },
     {
       "epoch": 9.076923076923077,
-      "grad_norm": 1.8668149709701538,
+      "grad_norm": 1.9303114414215088,
       "learning_rate": 7.897435897435898e-06,
-      "loss": 0.9712,
+      "loss": 0.8952,
       "step": 1180
     },
     {
       "epoch": 9.153846153846153,
-      "grad_norm": 2.042644739151001,
+      "grad_norm": 2.1960716247558594,
       "learning_rate": 7.794871794871796e-06,
-      "loss": 0.9407,
+      "loss": 0.8401,
       "step": 1190
     },
     {
       "epoch": 9.23076923076923,
-      "grad_norm": 1.967020869255066,
+      "grad_norm": 2.7040772438049316,
       "learning_rate": 7.692307692307694e-06,
-      "loss": 0.9457,
+      "loss": 0.8488,
       "step": 1200
     },
     {
       "epoch": 9.307692307692308,
-      "grad_norm": 2.147862672805786,
+      "grad_norm": 2.097287654876709,
       "learning_rate": 7.58974358974359e-06,
-      "loss": 0.9442,
+      "loss": 0.8544,
       "step": 1210
     },
     {
       "epoch": 9.384615384615385,
-      "grad_norm": 1.8528053760528564,
+      "grad_norm": 2.012044906616211,
       "learning_rate": 7.487179487179488e-06,
-      "loss": 0.9526,
+      "loss": 0.8679,
       "step": 1220
     },
     {
       "epoch": 9.461538461538462,
-      "grad_norm": 3.2000551223754883,
+      "grad_norm": 3.2651612758636475,
       "learning_rate": 7.384615384615386e-06,
-      "loss": 0.9465,
+      "loss": 0.8558,
       "step": 1230
     },
     {
       "epoch": 9.538461538461538,
-      "grad_norm": 2.259323835372925,
+      "grad_norm": 2.5597023963928223,
       "learning_rate": 7.282051282051282e-06,
-      "loss": 0.9503,
+      "loss": 0.8719,
       "step": 1240
     },
     {
       "epoch": 9.615384615384615,
-      "grad_norm": 2.4054858684539795,
+      "grad_norm": 2.756129503250122,
       "learning_rate": 7.17948717948718e-06,
-      "loss": 0.9274,
+      "loss": 0.8272,
       "step": 1250
     },
     {
       "epoch": 9.692307692307692,
-      "grad_norm": 3.4811408519744873,
+      "grad_norm": 2.7830724716186523,
       "learning_rate": 7.076923076923078e-06,
-      "loss": 0.943,
+      "loss": 0.854,
       "step": 1260
     },
     {
       "epoch": 9.76923076923077,
-      "grad_norm": 1.7080141305923462,
+      "grad_norm": 1.841320514678955,
       "learning_rate": 6.974358974358974e-06,
-      "loss": 0.9247,
+      "loss": 0.8283,
       "step": 1270
     },
     {
       "epoch": 9.846153846153847,
-      "grad_norm": 2.0476508140563965,
+      "grad_norm": 2.4447691440582275,
       "learning_rate": 6.871794871794872e-06,
-      "loss": 0.9194,
+      "loss": 0.8179,
       "step": 1280
     },
     {
       "epoch": 9.923076923076923,
-      "grad_norm": 2.149641990661621,
+      "grad_norm": 2.266535758972168,
       "learning_rate": 6.76923076923077e-06,
-      "loss": 0.9269,
+      "loss": 0.8368,
       "step": 1290
     },
     {
       "epoch": 10.0,
-      "grad_norm": 5.121323108673096,
+      "grad_norm": 4.2985944747924805,
       "learning_rate": 6.666666666666667e-06,
-      "loss": 0.9113,
+      "loss": 0.8112,
       "step": 1300
     },
     {
       "epoch": 10.0,
-      "eval_accuracy": 0.7819548872180451,
-      "eval_loss": 0.9149269461631775,
-      "eval_runtime": 0.7775,
-      "eval_samples_per_second": 171.06,
-      "eval_steps_per_second": 21.865,
+      "eval_accuracy": 0.8345864661654135,
+      "eval_loss": 0.8083619475364685,
+      "eval_runtime": 0.8269,
+      "eval_samples_per_second": 160.851,
+      "eval_steps_per_second": 20.56,
       "step": 1300
     },
     {
       "epoch": 10.076923076923077,
-      "grad_norm": 2.06109881401062,
+      "grad_norm": 1.931143045425415,
       "learning_rate": 6.564102564102565e-06,
-      "loss": 0.925,
+      "loss": 0.8378,
       "step": 1310
     },
     {
       "epoch": 10.153846153846153,
-      "grad_norm": 1.9137018918991089,
+      "grad_norm": 2.4910976886749268,
       "learning_rate": 6.461538461538463e-06,
-      "loss": 0.9657,
+      "loss": 0.8909,
       "step": 1320
     },
     {
       "epoch": 10.23076923076923,
-      "grad_norm": 2.0686280727386475,
+      "grad_norm": 2.3511204719543457,
       "learning_rate": 6.358974358974359e-06,
-      "loss": 0.9565,
+      "loss": 0.8715,
       "step": 1330
     },
     {
       "epoch": 10.307692307692308,
-      "grad_norm": 2.046623945236206,
+      "grad_norm": 2.0618457794189453,
       "learning_rate": 6.256410256410257e-06,
-      "loss": 0.918,
+      "loss": 0.833,
       "step": 1340
     },
     {
       "epoch": 10.384615384615385,
-      "grad_norm": 2.281343936920166,
+      "grad_norm": 2.4037184715270996,
       "learning_rate": 6.153846153846155e-06,
-      "loss": 0.9118,
+      "loss": 0.809,
       "step": 1350
     },
     {
       "epoch": 10.461538461538462,
-      "grad_norm": 2.694427728652954,
+      "grad_norm": 2.6920347213745117,
       "learning_rate": 6.051282051282051e-06,
-      "loss": 0.9377,
+      "loss": 0.8441,
       "step": 1360
     },
     {
       "epoch": 10.538461538461538,
-      "grad_norm": 2.3148765563964844,
+      "grad_norm": 2.5570709705352783,
       "learning_rate": 5.948717948717949e-06,
-      "loss": 0.911,
+      "loss": 0.8166,
       "step": 1370
     },
     {
       "epoch": 10.615384615384615,
-      "grad_norm": 2.595669746398926,
+      "grad_norm": 2.6978418827056885,
       "learning_rate": 5.846153846153847e-06,
-      "loss": 0.9146,
+      "loss": 0.8193,
       "step": 1380
     },
     {
       "epoch": 10.692307692307692,
-      "grad_norm": 2.136301279067993,
+      "grad_norm": 2.370861291885376,
       "learning_rate": 5.743589743589743e-06,
-      "loss": 0.9061,
+      "loss": 0.81,
       "step": 1390
     },
     {
       "epoch": 10.76923076923077,
-      "grad_norm": 3.0159363746643066,
+      "grad_norm": 3.260789394378662,
       "learning_rate": 5.641025641025641e-06,
-      "loss": 0.9365,
+      "loss": 0.8671,
       "step": 1400
     },
     {
       "epoch": 10.846153846153847,
-      "grad_norm": 2.1136507987976074,
+      "grad_norm": 2.27559494972229,
       "learning_rate": 5.538461538461539e-06,
-      "loss": 0.9232,
+      "loss": 0.8321,
       "step": 1410
     },
     {
       "epoch": 10.923076923076923,
-      "grad_norm": 1.713663101196289,
+      "grad_norm": 1.8184912204742432,
       "learning_rate": 5.435897435897436e-06,
-      "loss": 0.9344,
+      "loss": 0.8388,
       "step": 1420
     },
     {
       "epoch": 11.0,
-      "grad_norm": 4.04538631439209,
+      "grad_norm": 4.08992862701416,
       "learning_rate": 5.333333333333334e-06,
-      "loss": 0.9255,
+      "loss": 0.8335,
       "step": 1430
     },
     {
       "epoch": 11.0,
-      "eval_accuracy": 0.7894736842105263,
-      "eval_loss": 0.8905543088912964,
-      "eval_runtime": 0.7659,
-      "eval_samples_per_second": 173.659,
-      "eval_steps_per_second": 22.197,
+      "eval_accuracy": 0.8345864661654135,
+      "eval_loss": 0.7785491943359375,
+      "eval_runtime": 0.8255,
+      "eval_samples_per_second": 161.106,
+      "eval_steps_per_second": 20.592,
       "step": 1430
     },
     {
       "epoch": 11.076923076923077,
-      "grad_norm": 2.04194974899292,
+      "grad_norm": 2.0166661739349365,
       "learning_rate": 5.230769230769232e-06,
-      "loss": 0.9333,
+      "loss": 0.8471,
       "step": 1440
     },
     {
       "epoch": 11.153846153846153,
-      "grad_norm": 3.108344554901123,
+      "grad_norm": 3.2678654193878174,
       "learning_rate": 5.128205128205128e-06,
-      "loss": 0.9174,
+      "loss": 0.8298,
       "step": 1450
     },
     {
       "epoch": 11.23076923076923,
-      "grad_norm": 2.406233072280884,
+      "grad_norm": 2.545058488845825,
       "learning_rate": 5.025641025641026e-06,
-      "loss": 0.8948,
+      "loss": 0.7991,
       "step": 1460
     },
     {
       "epoch": 11.307692307692308,
-      "grad_norm": 2.4100501537323,
+      "grad_norm": 2.469082832336426,
       "learning_rate": 4.923076923076924e-06,
-      "loss": 0.9155,
+      "loss": 0.8256,
       "step": 1470
     },
     {
       "epoch": 11.384615384615385,
-      "grad_norm": 2.7117860317230225,
+      "grad_norm": 2.8207569122314453,
       "learning_rate": 4.820512820512821e-06,
-      "loss": 0.9075,
+      "loss": 0.8172,
       "step": 1480
     },
     {
       "epoch": 11.461538461538462,
-      "grad_norm": 2.0159695148468018,
+      "grad_norm": 2.142630100250244,
       "learning_rate": 4.717948717948718e-06,
-      "loss": 0.9338,
+      "loss": 0.8448,
       "step": 1490
     },
     {
       "epoch": 11.538461538461538,
-      "grad_norm": 3.280245304107666,
+      "grad_norm": 2.3974075317382812,
       "learning_rate": 4.615384615384616e-06,
-      "loss": 0.9243,
+      "loss": 0.8292,
       "step": 1500
     },
     {
       "epoch": 11.615384615384615,
-      "grad_norm": 3.1355690956115723,
+      "grad_norm": 3.240954875946045,
       "learning_rate": 4.512820512820513e-06,
-      "loss": 0.9185,
+      "loss": 0.8275,
       "step": 1510
     },
     {
       "epoch": 11.692307692307692,
-      "grad_norm": 3.0900094509124756,
+      "grad_norm": 3.3133740425109863,
       "learning_rate": 4.4102564102564104e-06,
-      "loss": 0.937,
+      "loss": 0.8569,
       "step": 1520
     },
     {
       "epoch": 11.76923076923077,
-      "grad_norm": 1.8758033514022827,
+      "grad_norm": 2.0835375785827637,
       "learning_rate": 4.307692307692308e-06,
-      "loss": 0.9052,
+      "loss": 0.8086,
       "step": 1530
     },
     {
       "epoch": 11.846153846153847,
-      "grad_norm": 2.0586955547332764,
+      "grad_norm": 2.20538067817688,
       "learning_rate": 4.2051282051282055e-06,
-      "loss": 0.8874,
+      "loss": 0.797,
       "step": 1540
     },
     {
       "epoch": 11.923076923076923,
-      "grad_norm": 2.0720062255859375,
+      "grad_norm": 2.1278247833251953,
       "learning_rate": 4.102564102564103e-06,
-      "loss": 0.9141,
+      "loss": 0.8301,
       "step": 1550
     },
     {
       "epoch": 12.0,
-      "grad_norm": 3.183523416519165,
+      "grad_norm": 3.4031107425689697,
       "learning_rate": 4.000000000000001e-06,
-      "loss": 0.8877,
+      "loss": 0.8062,
       "step": 1560
     },
     {
       "epoch": 12.0,
-      "eval_accuracy": 0.7894736842105263,
-      "eval_loss": 0.8749483823776245,
-      "eval_runtime": 0.7374,
-      "eval_samples_per_second": 180.372,
-      "eval_steps_per_second": 23.055,
+      "eval_accuracy": 0.8345864661654135,
+      "eval_loss": 0.7569367289543152,
+      "eval_runtime": 0.799,
+      "eval_samples_per_second": 166.458,
+      "eval_steps_per_second": 21.277,
       "step": 1560
     },
     {
       "epoch": 12.076923076923077,
-      "grad_norm": 2.0058720111846924,
+      "grad_norm": 2.181518793106079,
       "learning_rate": 3.897435897435898e-06,
-      "loss": 0.8829,
+      "loss": 0.7903,
       "step": 1570
     },
     {
       "epoch": 12.153846153846153,
-      "grad_norm": 2.2991676330566406,
+      "grad_norm": 2.4533674716949463,
       "learning_rate": 3.794871794871795e-06,
-      "loss": 0.9152,
+      "loss": 0.8228,
       "step": 1580
     },
     {
       "epoch": 12.23076923076923,
-      "grad_norm": 1.5903538465499878,
+      "grad_norm": 1.6895153522491455,
       "learning_rate": 3.692307692307693e-06,
-      "loss": 0.9149,
+      "loss": 0.8242,
       "step": 1590
     },
     {
       "epoch": 12.307692307692308,
-      "grad_norm": 1.7883615493774414,
+      "grad_norm": 2.1413424015045166,
       "learning_rate": 3.58974358974359e-06,
-      "loss": 0.9163,
+      "loss": 0.8378,
       "step": 1600
     },
     {
       "epoch": 12.384615384615385,
-      "grad_norm": 2.2841601371765137,
+      "grad_norm": 2.416987657546997,
       "learning_rate": 3.487179487179487e-06,
-      "loss": 0.8958,
+      "loss": 0.8088,
       "step": 1610
     },
     {
       "epoch": 12.461538461538462,
-      "grad_norm": 2.3814501762390137,
+      "grad_norm": 1.704318881034851,
       "learning_rate": 3.384615384615385e-06,
-      "loss": 0.8918,
+      "loss": 0.7953,
       "step": 1620
     },
     {
       "epoch": 12.538461538461538,
-      "grad_norm": 1.9848734140396118,
+      "grad_norm": 2.015375852584839,
       "learning_rate": 3.2820512820512823e-06,
-      "loss": 0.889,
+      "loss": 0.799,
       "step": 1630
     },
     {
       "epoch": 12.615384615384615,
-      "grad_norm": 1.7236778736114502,
+      "grad_norm": 1.851975917816162,
       "learning_rate": 3.1794871794871795e-06,
-      "loss": 0.8979,
+      "loss": 0.813,
       "step": 1640
     },
     {
       "epoch": 12.692307692307692,
-      "grad_norm": 3.340665102005005,
+      "grad_norm": 3.352241277694702,
       "learning_rate": 3.0769230769230774e-06,
-      "loss": 0.8695,
+      "loss": 0.763,
       "step": 1650
     },
     {
       "epoch": 12.76923076923077,
-      "grad_norm": 2.127927780151367,
+      "grad_norm": 2.4369592666625977,
       "learning_rate": 2.9743589743589746e-06,
-      "loss": 0.9323,
+      "loss": 0.8583,
       "step": 1660
     },
     {
       "epoch": 12.846153846153847,
-      "grad_norm": 1.8213707208633423,
+      "grad_norm": 1.7659847736358643,
       "learning_rate": 2.8717948717948717e-06,
-      "loss": 0.9178,
+      "loss": 0.8354,
       "step": 1670
     },
     {
       "epoch": 12.923076923076923,
-      "grad_norm": 2.0011963844299316,
+      "grad_norm": 1.8726952075958252,
       "learning_rate": 2.7692307692307697e-06,
-      "loss": 0.8872,
+      "loss": 0.7972,
       "step": 1680
     },
     {
       "epoch": 13.0,
-      "grad_norm": 3.812871217727661,
+      "grad_norm": 3.766446352005005,
       "learning_rate": 2.666666666666667e-06,
-      "loss": 0.9032,
+      "loss": 0.8141,
       "step": 1690
     },
     {
       "epoch": 13.0,
-      "eval_accuracy": 0.7969924812030075,
-      "eval_loss": 0.8698711395263672,
-      "eval_runtime": 0.7423,
-      "eval_samples_per_second": 179.165,
-      "eval_steps_per_second": 22.901,
+      "eval_accuracy": 0.849624060150376,
+      "eval_loss": 0.7535560131072998,
+      "eval_runtime": 0.8236,
+      "eval_samples_per_second": 161.479,
+      "eval_steps_per_second": 20.64,
       "step": 1690
     },
     {
       "epoch": 13.076923076923077,
-      "grad_norm": 2.8741540908813477,
+      "grad_norm": 3.104832649230957,
       "learning_rate": 2.564102564102564e-06,
-      "loss": 0.8842,
+      "loss": 0.7937,
       "step": 1700
     },
     {
       "epoch": 13.153846153846153,
-      "grad_norm": 2.3278818130493164,
+      "grad_norm": 2.46419095993042,
       "learning_rate": 2.461538461538462e-06,
-      "loss": 0.9131,
+      "loss": 0.822,
       "step": 1710
     },
     {
       "epoch": 13.23076923076923,
-      "grad_norm": 2.8419501781463623,
+      "grad_norm": 3.035752296447754,
       "learning_rate": 2.358974358974359e-06,
-      "loss": 0.8965,
+      "loss": 0.8127,
       "step": 1720
     },
     {
       "epoch": 13.307692307692308,
-      "grad_norm": 1.8506221771240234,
+      "grad_norm": 1.9470882415771484,
       "learning_rate": 2.2564102564102566e-06,
-      "loss": 0.8967,
+      "loss": 0.8151,
       "step": 1730
     },
     {
       "epoch": 13.384615384615385,
-      "grad_norm": 2.6166839599609375,
+      "grad_norm": 2.086946964263916,
       "learning_rate": 2.153846153846154e-06,
-      "loss": 0.8785,
+      "loss": 0.7773,
       "step": 1740
     },
     {
       "epoch": 13.461538461538462,
-      "grad_norm": 4.287515640258789,
+      "grad_norm": 3.7044293880462646,
       "learning_rate": 2.0512820512820513e-06,
-      "loss": 0.914,
+      "loss": 0.8276,
       "step": 1750
     },
     {
       "epoch": 13.538461538461538,
-      "grad_norm": 2.516889810562134,
+      "grad_norm": 2.735135555267334,
       "learning_rate": 1.948717948717949e-06,
-      "loss": 0.9286,
+      "loss": 0.8462,
       "step": 1760
     },
     {
       "epoch": 13.615384615384615,
-      "grad_norm": 1.8332946300506592,
+      "grad_norm": 2.065619468688965,
       "learning_rate": 1.8461538461538465e-06,
-      "loss": 0.8995,
+      "loss": 0.8024,
       "step": 1770
     },
     {
       "epoch": 13.692307692307692,
-      "grad_norm": 2.2418551445007324,
+      "grad_norm": 2.519625663757324,
       "learning_rate": 1.7435897435897436e-06,
-      "loss": 0.8818,
+      "loss": 0.7813,
       "step": 1780
     },
     {
       "epoch": 13.76923076923077,
-      "grad_norm": 1.794832706451416,
+      "grad_norm": 1.8556421995162964,
       "learning_rate": 1.6410256410256412e-06,
-      "loss": 0.9044,
+      "loss": 0.8218,
       "step": 1790
     },
     {
       "epoch": 13.846153846153847,
-      "grad_norm": 3.0142152309417725,
+      "grad_norm": 3.1282896995544434,
       "learning_rate": 1.5384615384615387e-06,
-      "loss": 0.8826,
+      "loss": 0.7891,
       "step": 1800
     },
     {
       "epoch": 13.923076923076923,
-      "grad_norm": 2.5891315937042236,
+      "grad_norm": 2.363374948501587,
       "learning_rate": 1.4358974358974359e-06,
-      "loss": 0.8387,
+      "loss": 0.7341,
       "step": 1810
     },
     {
       "epoch": 14.0,
-      "grad_norm": 5.37412691116333,
+      "grad_norm": 4.762740612030029,
       "learning_rate": 1.3333333333333334e-06,
-      "loss": 0.9001,
+      "loss": 0.8172,
       "step": 1820
     },
     {
       "epoch": 14.0,
-      "eval_accuracy": 0.7819548872180451,
-      "eval_loss": 0.8673797845840454,
-      "eval_runtime": 0.7642,
-      "eval_samples_per_second": 174.027,
-      "eval_steps_per_second": 22.244,
+      "eval_accuracy": 0.8270676691729323,
+      "eval_loss": 0.7531744837760925,
+      "eval_runtime": 0.8266,
+      "eval_samples_per_second": 160.91,
+      "eval_steps_per_second": 20.567,
       "step": 1820
     },
     {
       "epoch": 14.076923076923077,
-      "grad_norm": 1.8213236331939697,
+      "grad_norm": 1.8959217071533203,
       "learning_rate": 1.230769230769231e-06,
-      "loss": 0.9047,
+      "loss": 0.8201,
       "step": 1830
     },
     {
       "epoch": 14.153846153846153,
-      "grad_norm": 1.8006333112716675,
+      "grad_norm": 2.1285548210144043,
       "learning_rate": 1.1282051282051283e-06,
-      "loss": 0.8768,
+      "loss": 0.7906,
       "step": 1840
     },
     {
       "epoch": 14.23076923076923,
-      "grad_norm": 2.691574811935425,
+      "grad_norm": 2.916303873062134,
       "learning_rate": 1.0256410256410257e-06,
-      "loss": 0.8757,
+      "loss": 0.7793,
       "step": 1850
     },
     {
       "epoch": 14.307692307692308,
-      "grad_norm": 5.015848636627197,
+      "grad_norm": 2.9460341930389404,
       "learning_rate": 9.230769230769232e-07,
-      "loss": 0.8734,
+      "loss": 0.773,
       "step": 1860
     },
     {
       "epoch": 14.384615384615385,
-      "grad_norm": 2.5233821868896484,
+      "grad_norm": 2.4907443523406982,
       "learning_rate": 8.205128205128206e-07,
-      "loss": 0.8787,
+      "loss": 0.7887,
       "step": 1870
     },
     {
       "epoch": 14.461538461538462,
-      "grad_norm": 2.1718924045562744,
+      "grad_norm": 2.431607961654663,
       "learning_rate": 7.179487179487179e-07,
-      "loss": 0.8767,
+      "loss": 0.7886,
       "step": 1880
     },
     {
       "epoch": 14.538461538461538,
-      "grad_norm": 3.0364015102386475,
+      "grad_norm": 3.005627155303955,
       "learning_rate": 6.153846153846155e-07,
-      "loss": 0.873,
+      "loss": 0.7789,
       "step": 1890
     },
     {
       "epoch": 14.615384615384615,
-      "grad_norm": 2.5152034759521484,
+      "grad_norm": 2.981405735015869,
       "learning_rate": 5.128205128205128e-07,
-      "loss": 0.9096,
+      "loss": 0.8262,
       "step": 1900
     },
     {
       "epoch": 14.692307692307692,
-      "grad_norm": 1.819096565246582,
+      "grad_norm": 1.9199621677398682,
       "learning_rate": 4.102564102564103e-07,
-      "loss": 0.892,
+      "loss": 0.8015,
       "step": 1910
     },
     {
       "epoch": 14.76923076923077,
-      "grad_norm": 3.512732982635498,
+      "grad_norm": 3.6987714767456055,
       "learning_rate": 3.0769230769230774e-07,
-      "loss": 0.8937,
+      "loss": 0.7998,
       "step": 1920
     },
     {
       "epoch": 14.846153846153847,
-      "grad_norm": 2.917677879333496,
+      "grad_norm": 3.5848920345306396,
       "learning_rate": 2.0512820512820514e-07,
-      "loss": 0.9231,
+      "loss": 0.8552,
       "step": 1930
     },
     {
       "epoch": 14.923076923076923,
-      "grad_norm": 2.0683395862579346,
+      "grad_norm": 2.4082224369049072,
       "learning_rate": 1.0256410256410257e-07,
-      "loss": 0.8613,
+      "loss": 0.759,
       "step": 1940
     },
     {
       "epoch": 15.0,
-      "grad_norm": 4.704519271850586,
+      "grad_norm": 4.4640116691589355,
       "learning_rate": 0.0,
-      "loss": 0.8842,
+      "loss": 0.7896,
       "step": 1950
     },
     {
       "epoch": 15.0,
-      "eval_accuracy": 0.7894736842105263,
-      "eval_loss": 0.8550169467926025,
-      "eval_runtime": 0.7718,
-      "eval_samples_per_second": 172.329,
-      "eval_steps_per_second": 22.027,
+      "eval_accuracy": 0.8045112781954887,
+      "eval_loss": 0.7388833165168762,
+      "eval_runtime": 0.8692,
+      "eval_samples_per_second": 153.016,
+      "eval_steps_per_second": 19.558,
       "step": 1950
     },
     {
       "epoch": 15.0,
       "step": 1950,
       "total_flos": 1.5658365504595968e+17,
-      "train_loss": 0.6299933981284117,
-      "train_runtime": 92.772,
-      "train_samples_per_second": 167.184,
-      "train_steps_per_second": 21.019
+      "train_loss": 0.923483537771763,
+      "train_runtime": 151.8082,
+      "train_samples_per_second": 102.168,
+      "train_steps_per_second": 12.845
     }
   ],
   "logging_steps": 10,